Background: Each year, millions of Americans are hospitalized for suspected infection and receive antibiotics. Current guidelines strongly urge broad-spectrum antibiotics be delivered within 1 hour in the most-life- threatening cases: severe sepsis and septic shock. However, unlike other time-sensitive medical emergencies, there is no objective standard for the diagnosis of severe sepsis. Yet, several government initiatives incentivize rapid antibiotic treatment for severe sepsis. There are growing concerns that this intense focus on early antibiotics will increase overall antibiotic use, contributing to antimicrobial resistance. This R01 will measure trade-offs associated with accelerating time-to-antibiotics to inform both individual bedside decision-making and federal policy. Specific Aims: (A1) Hospital antibiotic prescribing behavior: When hospitals accelerate timing of antibiotic delivery for severe sepsis, how often is this change associated with increasing antibiotic use and broader spectrum of coverage among all-comers with potential infection? (A2) Individual harms/benefits: For an individual patient, we will quantify the: (a) risks of harm associated with very short courses of antibiotics (allergic reaction, renal or liver injury, cytopenias, C. difficile infection, isolation of new resistant bacteria, and mortality); (b) mortality benefit associated with earlier delivery of antibiotics. A3) Hospital net harms/benefits: Based on temporal changes in antibiotic prescribing (Aim 1), patient-level harms/benefits (Aim 2), and hospital prevalence of severe sepsis and septic shock, we will estimate the net benefits and harms associated with temporal changes in antibiotic prescribing with simulation. Anticipated Impact: This work will quantify the prescribing trade-offs associated with accelerating time-to- antibiotics, the individual patient-level harms of very short courses of antibiotics, and the benefit of faster antibiotic delivery, across subgroups of patients. Finally, we will quantify the net benefits and harms of accelerating time-to-antibiotics across different types of hospitals with varying sepsis prevalence and magnitude of negative externalities associated with accelerating time-to-antibiotics. Unique Features and Innovation: Using granular, patient-level data from Kaiser Permanente Northern California (KPNC) and nationwide Veterans Affairs (VA), and a novel measure of antimicrobial coverage spectrum (Spectrum Score), we will evaluate the holistic benefits and harms of faster antibiotic timing, making summary judgements about the balance of benefits straightforward and informative. Project Methods: We will examine all KNPC and VA hospitalizations (2013-2018) admitted with potential infection. We will assess hospital-specific temporal trends in antibiotic prescribing patterns using multilevel models. We will assess patient-level harms/benefit using matching and regression. We will estimate the plausible range of trade-offs that result from accelerating time-to-antibiotics with simulation.